JP2001282539A - Method and device for structurizing concept and device provided with concept structure - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce the input cost of a user himself (or herself) and to improve the expression accuracy of a concept structure map. SOLUTION: Physical stimulus generated by a physical stimulus generation means 10 is outputted from a physical stimulus output means 12. A user receiving the physical stimulus inputs a user's verbal response corresponding to the physical stimulus to an autonomous input means 14 and an automatic measuring means 16 automatically measures a user's non-verbal response. A concept map generation means 18 generates an initial concept structure map expressing relation between the physical stimulus applied to the user and the user's verbal response corresponding to the physical stimulus. A correction coefficient determination means 20 operates a correction coefficient on the basis of the non-verbal response information and an automatic concept map correction means 22 corrects relation between both the information data by changing a distance between the elements of the initial concept structure map on the basis of the correction coefficient.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

FIELD OF THE INVENTION The present invention relates to a method and apparatus for structuring a concept, and to an apparatus having a concept structure.
By generating a conceptual structure map representing the relationship between a physical stimulus to a human and a human first response to the physical stimulus, and correcting the generated conceptual structure map based on the simultaneously generated second response, The present invention relates to a concept structuring method and apparatus capable of obtaining a conceptual structure map suitable for an emotional structure, and an apparatus provided with the obtained conceptual structure map.

[0002]

2. Description of the Related Art As a technique for obtaining a conceptual structure map adapted to an individual's emotional structure, a technique (EVIDI) for visualizing the meaning and impression of an image while interactively acquiring the relation between the user and the user.
I [Environment for Visualiz
ING INDIVIDUAL DIFFERENCES
in Impressions of Images) [ht
tp: //ccc.aist-nara.ac.jp/systems/evidii/]) is known. In this technology, when correcting the conceptual structure map, it relies on human conscious correction, and it is not corrected by the event that occurred simultaneously when creating the conceptual structure map,
The map cannot be automatically corrected.

In addition, a technique for organizing semantic distances between knowledge on a spatial map and creating a map by self-reporting the importance of conceptual elements by a user (one for supporting systematization of emergent knowledge). Method [The 15th Meeting of the Society of Instrument and Control Engineers, 15th meeting of the System Engineering Subcommittee, Sample "Ideal support technology", P39-P4
6, 1994]), and a technique in which the user directly corrects the relationship strength between conceptual elements on a space map using a hand icon (Japanese Patent Laid-Open No. 11-120162), is also known. The map cannot be automatically corrected because it is not corrected by an event that occurred simultaneously when the conceptual structure map was created.

That is, in each of the above-mentioned conventional techniques, the user needs to newly input information for correcting the strength of the relationship between elements (between concept elements) after inputting the first information for generating a map. Therefore, the time load of map generation increases, and the strength of the relationship between elements is corrected depending on the user's autonomous and conscious input.
There is a problem that the labor load for generating the map increases.

In addition, since the correction is performed depending on the relatively unclear language-based response information, there is a limit in the expression accuracy of the conceptual structure map, and it is difficult to generate a highly accurate conceptual structure map. Is a problem.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and has a method and an apparatus for structuring a concept in which the input cost (time load, labor load, etc.) of the user himself is reduced, and the generated method. It is a first object of the present invention to provide an apparatus having a conceptual structure.

[0007] The present invention also considers the uncertainty (language and ambiguity of the user) of the linguistic information, the difference in the definition (interpretation of the meaning, the range of the object, etc.) for the language, and the restraint by the unarranged value standard. It is a second object of the present invention to provide a method and an apparatus for structuring a concept, which can cover and improve the expression accuracy of the concept structure, and an apparatus having the generated concept structure.

[0008]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention provides a conceptual structure representing a relationship between a physical stimulus to a human and a first response of the human to the physical stimulus. A second response that occurred with the
For example, the second that occurred simultaneously with the application of the physical stimulus
Or a second response generated within a short period of time after the physical stimulus is given to obtain a conceptual structure suitable for the emotional structure of the individual. This conceptual structure can be represented visually on a map.

According to the present invention, a conceptual structure representing the relationship between a physical stimulus and a first response of a human is defined as a second response generated when a physical stimulus is given to a human, for example, presentation of a physical stimulus. Correction is performed based on the second response that has occurred at the same time.
Therefore, the time load of generating the conceptual structure can be reduced.

The conceptual structure representing the relationship between the physical stimulus and the first response of the human may be generated as the physical stimulus is given to the human, or may be generated simultaneously with the presentation of the physical stimulus. In addition, a conceptual structure generated in advance before the presentation of the physical stimulus may be used.

In the present invention, if the concept structure is automatically corrected by automatically measuring the second response generated when a physical stimulus is given to a human, the concept structure can be generated. Labor load can be reduced.

In the present invention, the first response can be one of an autonomous response and a non-autonomous response, and the second response can be the other of the autonomous response and the non-autonomous response. At this time, the non-autonomous response may be a speech measurement value, a physiological measurement value, or both a speech measurement value and a physiological measurement value.

Further, the first response may be one of a verbal response and a non-verbal response, and the second response may be the other of a verbal response and a non-verbal response. At this time, the non-verbal response can be a speech measurement, a physiological measurement, or both a speech measurement and a physiological measurement.

That is, according to the present invention, a conceptual structure based on an autonomous response is corrected based on a nonautonomous response generated when a physical stimulus is given to a human, or a conceptual structure based on the nonautonomous response is corrected. Based on the concept structure based on the autonomous response generated when a physical stimulus is given to a human, the concept structure based on a linguistic response is corrected with a physical stimulus given to a human. Correction based on the nonverbal response that occurred during the simulation, or correction of the conceptual structure based on the nonverbal response based on the linguistic response that occurred when a physical stimulus was given to a human. can do.

Further, according to the present invention, environmental information around a human may be simultaneously measured as a physical stimulus is applied to the human, and the environmental information may be further added and corrected.

In the present invention, the conceptual structure is defined as, for example, 2
Initially generated with a three-dimensional, three-dimensional or other multi-dimensional map, and the initially generated conceptual structure is a multi-dimensional measurement obtained from a human (eg, a user), preferably a multi-dimensional measurement other than a language system Based on the values (non-verbal measurements)
Preferably, the correction is made automatically. In actual human communication, the non-verbal (non-verbal) response is used to estimate the conceptual structure of the opponent, so the information required for correction is obtained at the initial input for generating the conceptual structure.
It is preferable to acquire by simultaneously measuring non-verbal (non-verbal) responses (“between”, facial expressions, eyes, heartbeats, etc.) that the user unconsciously sends.

The conceptual structure obtained by the present invention can be applied to various devices such as an emotion-adaptive interior design image output system, a situation-adaptive vehicle interior information providing system, a pet robot, a care robot, and a counseling robot. .

[0018]

Embodiments of the present invention will be described below in detail with reference to the drawings. In the first embodiment, a conceptual structure map is generated based on an autonomous response, and the generated conceptual map is corrected based on a non-autonomous response.

As shown in FIG. 1, the first embodiment is
The apparatus includes a physical stimulus generation unit 10 and a physical stimulus output unit 12 that outputs a physical stimulus generated by the physical stimulus generation unit 10. The physical stimulus output means 12 provides a physical stimulus to a human user.

The user who has received the physical stimulus inputs a verbal (verbal) response, which is the user's first response to the physical stimulus, to the autonomous input means 14. This linguistic response is an autonomous response and is input to the concept map generating means 18 together with the time. Also, the physical stimulus information from the physical stimulus generating means 10 is input to the concept map generating means 18 together with the time information at which the physical stimulus was given. When the physical stimulus is presented to the user and verbal response information is input, the concept map generation unit 18 refers to the time information and displays the physical stimulus to the user and the human verbal response to the physical stimulus as shown in FIG. Generate an initial conceptual structure map representing the relationship with.

At the same time as the physical stimulus is given to the user, the non-verbal (non-verbal) response of the human is automatically measured by the automatic measuring means (automatic input means) 16. This non-verbal (non-verbal) response is a non-autonomous response, and is input to the correction coefficient determining means 20 as time-series measurement data together with measurement time information. The physical coefficient information from the physical stimulus generating means 10 is input to the correction coefficient determining means 20 together with the time at which the physical stimulus was given. The correction coefficient determining means 20 determines the time at which the verbal response information is input to the autonomous response means 14 based on the non-verbal response information, the physical stimulus information, and the time information (the time at which the physical stimulus is given to the user); That is, a correction coefficient at the same time as the time when the initial conceptual structure map is generated is calculated. Accordingly, a correction coefficient for correcting the conceptual structure map generated by the verbal response information is determined by the non-verbal response information. Further, since the non-verbal response information is generated at the same time as the verbal response information for generating the conceptual structure map, the conceptual structure map can be accurately corrected.

The initial concept structure map generated by the concept map generation means 18 is input to the concept map automatic correction means 22. Also, the correction coefficient at the time corresponding to the time when the initial conceptual structure map is generated is input to the concept map automatic correction unit 22 from the correction coefficient determination unit 20. As shown in FIG. 3, the concept map automatic correction unit 22 corrects the relationship between information by changing the distance between elements of the initial concept structure map using a correction coefficient. The corrected conceptual structure map is output from the result output means 24.

Next, each part of the above embodiment will be described in detail.

The physical stimulus generating means 10 comprises a computer or the like in which physical stimulus information is stored.
2 can be constituted by an engineering output medium such as a monitor, a speaker, a printer, and the like. It should be noted that not only the output from the output medium but also the existence itself of the scenery and the substance may be used as the physical stimulus. In the case where the existence itself of a scenery, a substance, or the like is used as a physical stimulus, the physical stimulus generation unit 10 and the physical stimulus output unit 12 are unnecessary. Also, the physical stimulus
Instead of intentionally prepared physical stimulus, any physical stimulus may be used. The physical stimulus is not limited to a stimulus that stimulates visual and auditory senses, but may be a stimulus that stimulates the human senses (for example, touch, taste, and smell). This physical stimulus may be used alone or in combination.

The output from the result output means 24 may be fed back and used as a physical stimulus.

The autonomous input means 14 is for inputting a response (autonomous response) conscious of the user, and is constituted by a device capable of inputting verbal information such as a keyboard, a mouse, a touch panel, a microphone, and a camera. Autonomous input means 1
4, verbal response information such as keywords, themes, summaries, sentences, conversations, etc. is autonomously input. When inputting the response information, the response information may be selected and input from a plurality of verbal information such as keywords, sentences, and conversation sentences registered in the computer in advance.

[0027] In addition to the input by the computer, an element on which verbal response information such as a keyword, a sentence, or a conversation is described is arranged on a screen, a board, or a sheet of paper. (By changing the position of the element and the distance between the elements).

It is also possible for the user not to directly input, but to automatically input, for example, from a conversation through voice processing and natural language processing.

Further, necessary parts can be complemented and input by using general data stored in advance.
Without using the autonomous input means 14, the concept map generating means 1
8 or the concept map automatic correction means 22 may store a basic concept structure map generated in advance. When the basic concept structure map is stored in the concept map automatic correction unit 22, the concept map generation unit 18 is also unnecessary.

The automatic measuring means 16 measures a response (non-autonomous response) generated without the user's consciousness.
It can be composed of a measuring device that measures physiological data and speech data. Physiological data includes skin temperature (face, fingertips, body surface, etc.), heartbeat, blink, respiration, brain wave,
SCR (skin electrical response), myoelectric potential, pupil diameter, state of saliva and blood, and the like can also be measured. Regarding the utterance data measured by the automatic measuring means 16, data other than words such as "between", intonation (eg, confidence), volume (eg, emotion), speed (eg, tension), etc., can be measured. Good.

A change in facial expression, a complexion, a gesture, a habit such as poor extortion may be included in the measurement object. If the time data is also measured at the same time, the occurrence order of actions and the like can be accumulated as data. In addition to personal information of the user, information of others and environmental information (temperature and humidity, brightness, color, odor, weather, brightness, location, etc.) may be included in the measurement target and measured simultaneously.

When incorporated into a device that involves operations such as automobiles, the direction of operation, the amount of operation, the speed, road conditions (positions of pedestrians, conditions of surrounding vehicles), vehicle driving data by ITS and the like are also used as environmental information. It may be included in the measurement target.

The correction coefficient determining means 20 can determine a correction coefficient by weighting a plurality of measured values and integrating indexes. Not only the measured value itself, but also a deviation between the measured value and the average value can be used as a correction coefficient. Therefore, the correction coefficient can be determined by reflecting the measured values before and after the physical stimulus is applied, without using only the value measured simultaneously with the application of the physical stimulus. That is,
The correction coefficient can be determined based on a measurement value generated when a physical stimulus is given to a human.

As shown in FIG. 2, based on the physical stimulus information from the physical stimulus generating means 10 and the verbal response information, the conceptual map generating means 18 outputs information (a plurality of physical stimulus information and a plurality of verbal response information). Information) Create a conceptual structure map that visualizes the overall structure of the concept by spatially arranging the individual. In the conceptual structure map, the shorter the distance between information (between elements), the deeper the relationship between information.

In this conceptual structure map, a plurality of physical stimulus information (voice, still image, moving image, photograph, etc.) and a plurality of verbal response information (keyword, theme, summary, etc.) are used as a plurality of elements. Elements can be arranged in an unlimited number.

Each element on the conceptual structure map can include not only sensory information of the user but also scene information (age, gender, experience, etc.) other than sensory information.

Each element on the conceptual structure map can store information such as strength, importance, and comfort as properties inside itself.

An object such as a computer program may be arranged on the conceptual structure map. The conceptual structure map is a multidimensional space, and the number of dimensions can be increased as much as possible.

A plurality of conceptual structure maps may exist. A different conceptual structure map may be used for each theme, or some conceptual structure maps may be summarized to hierarchically generate a higher-level conceptual structure map.

A large number of user conceptual structure maps may be generated simultaneously, and by activating one or more elements of the conceptual structure map, the activation is diffused to surrounding elements according to a specific function. Can also.

The automatic concept map correcting means 22 may perform the correction starting from the initial state of the concept map stored in advance, irrespective of the initial concept structure map input from the concept map generating means 18.

Further, the correction history can be accumulated and reused at the time of the next correction, and the conversion function used for correction, the number of correction coefficients, the correction timing, and the like can be arbitrarily set.

For the non-linear part of the conversion function, an exponential function represented by the following equation (6) can be used, but any function other than the exponential function can be used according to the purpose.

By using cluster analysis and the like in addition to the multidimensional scale construction method described below, a conceptual structure map summarizing semantically similar concepts can be created, and the conceptual structure map when visualized Interpretation of maps becomes easy.

Instead of the multidimensional scaling method, a quantum / electromagnetic method or a dynamic method such as a wave function shown in the following equation (1) can be applied. In addition, a neutral network, fuzzy logic, chaotic interpretation, and the like can be added.

[0046]

(Equation 1) From the result output means 24, a conceptual structure map in which only necessary elements are selected can be output.

The conceptual structure map is not only a visual display,
For example, by selecting an element on the conceptual structure map, it can be expressed as voice, mechanical movement, or execution of a program.

In addition to the concept structure map of the individual user, a comparison with the concept structure map of a plurality of users, a difference with the average concept structure map, and the like can be output.

The corrected conceptual structure map can be incorporated into another system via a network or a recording medium, for example, so that the personality of the user can be immediately reflected. Details of an application example of the conceptual structure map will be described later.

According to the present embodiment, the conceptual structure map is corrected based on the non-verbal response which is the second response generated simultaneously with the presentation of the physical stimulus.
A conceptual structure that matches the emotional structure of the individual can be obtained.

Further, a conceptual structure suitable for an individual's emotional structure can be obtained from utterance measurement values (interval, intonation, volume, speed, etc.).

Surrounding environment information (other information, temperature, humidity,
Correction by lightness, darkness, smell, etc.) can further adapt to the emotional structure of the individual.

Further, the non-verbal response (for example, “between”,
Expression, gaze, heart rate, etc.)
The concept map automatic correcting means automatically corrects the conceptual structure map based on the measurement information of the automatic measuring means without requesting a new response from the user. It is possible to improve the expression accuracy of the conceptual structure.

Next, FIG. 4 shows a second embodiment of the present invention. The second embodiment differs from the first embodiment only in that a conceptual structure map is generated based on a non-autonomous response and the generated conceptual structure map is corrected based on an autonomous response. Therefore, in FIG. 4, the portions corresponding to FIG.

Next, a description will be given of a third embodiment in which the present invention is applied to a system for outputting a design image adapted to an emotion using the first embodiment. This embodiment is an emotion-adaptive interior design image output system for supporting communication between a customer and a sales staff in a home sales situation.

FIG. 5 shows an apparatus for generating a conceptual structure map of this system. Since the apparatus has the same configuration as that of the apparatus shown in FIG. 1, parts corresponding to those in FIG. The physical stimulus generation means (not shown) reads a photograph related to the interior from a specialized magazine or the like with a scanner, captures the photograph in a personal computer, collects images for evaluating the interior as physical stimuli, and generates a database (D
B) is generated.

In the physical stimulus output means 12, as shown in the flow chart of FIG. 6, an evaluation image taken in the image database (DB) of the personal computer is selected, the evaluation image is presented on the screen, and the image is presented to the user. Is input as well as an impression word for the image (S10 to S16).

As shown in FIG. 7, the evaluation screen composed of the interior-related images 30 is displayed on the display screen of the personal computer and presented to the user as a physical stimulus. The user operates the autonomous input unit 14 such as a mouse to select an impression word that matches his / her expression from the display unit 32 on which the emotion expression word and the design expression word (impression word) are displayed on the screen. Enter verbal response information.

As the emotion expression words, “lively”,
Use languages such as "energetic", "energetic", "perky", "relaxed", "slow", "calm", "relaxed", "easy" As design expression words, “luxurious”, “warm”, “hard”, “dark”, “has a contrast”, “new”, “simplicity”, and the like can be used.

The automatic measuring means 16 measures the like / dislike result (preference for each image) and the judgment time of like / dislike as an action index, and a keyword (this keyword is predetermined for an emotion expression word and a design expression word) as a language index. Is measured), heart rate,
The electrogastrogram and palm blood flow as the skin blood flow are measured and input to each of the judgment response time database, image preference database, word use frequency database, and whole body activity coefficient database of the measurement database section of the correction coefficient determination means 20. I do. The physiological index is standardized and converted into a standardized amount (physiological score described later).

If there is no impression word sufficiently expressed (S18), the impression word is added to the language database to be used and selected (S20, S16).

When the impression words have been selected up to the upper limit of the number of evaluation images (S22), the evaluation images and the impression words are displayed on the screen as shown in FIG. Thereby, the user arranges the evaluation image and the impression word while adjusting the positions and distances of the plurality of elements including the evaluation image and the impression word by operating the mouse, and generates the initial conceptual structure map represented by the plurality of elements. Generate.

When the generation of the initial conceptual structure map is completed,
The personal computer acquires the coordinates of each element of the placed evaluation image and impression word, and calculates the distance between the elements (S
26).

The correction coefficient determining means 20 calculates a correction coefficient (distance adjustment coefficient) for adjusting the distance between the elements using the language index, action index, and physiological index input to the measurement data section (S28). ), The calculated correction coefficient is input to the conceptual map automatic correction means 22 as a correction coefficient corresponding to the physical stimulus information input from the physical quantity generation means.

Since the initial concept structure map represented by the initial arrangement of the language indices is input to the concept map automatic correcting means 22 from the concept map generating means 18, the distance between the evaluation image and the impression word (corresponding to the correction coefficient). (Including the distance between the emotion expression word and the design expression word), determine the optimal arrangement of the evaluation image and the impression word by multidimensional scaling, output the corrected conceptual structure map by rearranging the elements Do (S
30-S34). The corrected concept structure map is stored in the user concept database.

FIG. 9 schematically shows the initial conceptual structure map, a correction coefficient (distance adjustment coefficient) for adjusting the distance between elements using the language index, the behavior index, and the physiological index, and the corrected conceptual structure map. Shown in In FIG. 9, reference numeral 40 denotes an initial conceptual structure map, and reference numeral 50 denotes a corrected conceptual structure map.

Next, the scoring of the physiological index will be described in more detail. Hereinafter, in order to unify the direction of the change of the physiological index, the sympathetic nervous activity is defined as a plus direction, and the parasympathetic nervous activity is defined as a minus direction.

Regarding the heart rate scoring, using the following equation (2) for standardization, the change amount of the heart rate from the resting state is obtained, and the standard deviation at the resting state becomes 1 Standardized for each user as follows. Then, the value of the standardization of the equation (2) is directly used as the heart rate score.

[0069]

(Equation 2) Regarding scoring of the electrogastrogram, the amount of shift of the center frequency when the physical stimulus is presented from the original center frequency (0.05 Hz) of the electrogastrogram is calculated using the standard deviation (0. 0
(25 Hz) is standardized for each individual user so as to be 1.

Next, the parasympathetic nervous activity is enhanced (negative direction: maximum -1) as the central frequency at the time of stimulation is closer to 0.05 Hz, and 0.02 at 0.05 ± 0.025 Hz (standard deviation).
Scoring is performed according to the following equation (3) so that the movement of 5 Hz or more increases the sympathetic nervous activity (+ direction).

[0071]

(Equation 3) Regarding the scoring of palm blood flow, the change in blood flow from resting is standardized for each user according to the following equation (4) so that the magnitude of blood flow at rest is 1. In this case, when the blood flow increased, the parasympathetic nervous activity was increased, and when the blood flow decreased, the sympathetic nervous activity was increased.

[0072]

(Equation 4) Next, palm blood flow is a peripheral autonomic nervous activity, and its effect on the whole body is not considered to be as great as heart rate or EMG. did.

Examples in which physiological changes are scored according to the above equations (2) to (4) are shown in FIGS. 10 and 11, and the equation for total scoring is shown in equation (5). FIG. 12 shows the obtained scores for each user. FIG. 13 shows an example of the measurement result of the preference judgment time.

[0074]

(Equation 5) Also, the above-described multidimensional scaling method (MDS: Multi-Dimensional Scaling, Tomio Hayashi, “Multidimensional Scaling Analysis Method” Science, 1982) calculates the dissimilarity between element i and element j by the following (6). _Assuming that δ _ij is represented by the equation, and the Euclidean distance in the multidimensional space between the point representing the element i and the point representing the element j is d _ij, as shown in the following equation (7), The position (coordinates) of a point in a multidimensional space is determined so that the distance between elements having a smaller degree becomes smaller.

If δ _ij > δ _kl , d _ij ≧ d _kl, that is, the distance is adjusted so that the point-to-point distance increases monotonically with the dissimilarity, and the point-to-point distance and the dissimilarity are determined by the steepest descent method. The configuration (coordinates) that minimizes the degree of deviation (stress: S) from the monotonically increasing relationship with the above is obtained.

[0076]

(Equation 6) In the above (6), [delta] _ij denotes the dissimilarity between element i and element _{j, exp (-w i · w} j) is the weighting for the distance matrix, represent elements of keywords or evaluate image k Here, w _k is the frequency of use of impression words when k is a keyword, w _k is the like / dislike determination time when k is an evaluation image, t is the number of dimensions of the multidimensional scaling method, and x _kl is the element k
Is a coordinate value in the dimension 1 of, and P _ij is a whole body activity coefficient, that is, a correction coefficient.

As shown in FIG. 14, the conceptual structure map includes a plurality of individual conceptual structure maps 50A and one general expression representing a general relationship between design factors (design requirements) and design images (rooms, etc.). A conceptual structure map 50B is created. This design image can be collected from catalogs, commercial CDs, magazines, and the like.

As shown in FIGS. 14 and 15, a plurality of conceptual structure maps 50A are stored in the emotion-design language distance database 60 of the emotion-adaptive interior design image output system, and general relations are shown. The conceptual structure map 50B that represents the emotion-compatible interior design image is stored in the design-visual image pattern distance database 62 of the output system.

In this system, when a desired emotion requirement (emotional state) regarding a room is input, the input emotion requirement using the emotion-design language distance database 60 is converted into a design requirement (design factor). You. Also,
Design-visual image pattern distance database 6
Design requirements converted using 2 (design factors)
Is converted into a design image, and candidates for an interior design that match the input emotional requirements are output. as a result,
A result adapted to the personal characteristics of the customer is output, and even if the same emotion information is input, the output result differs depending on the person. In addition, images can be visualized simply by expressing one's feelings without using difficult design terminology. When the indoor image is finally narrowed down, the basis is scientific, quantitative and clear.

Next, a fourth embodiment in which the conceptual structure map generated according to the present invention is applied as a brain of a robot will be described.

The conceptual structure map generated by the present invention can be an effective model as a mechanism of human memory. Elements placed on this conceptual structure map are:
It can be compared as a memory trace, and the distance between each element in the multidimensional space represents the similarity of the lexical meaning. For example, when a certain element (memory trace) is activated, it can be made to function so that neighboring elements are sequentially activated according to a specific function. With this feature,
Difficult with conventional robot technology such as association and creation,
Higher-order functions including associative and creative functions unique to humans can be realized.

Further, in this conceptual structure map, a flexible learning function can be realized by adding an element to be arranged in the conceptual structure map as needed and newly forming a distance relationship with an existing element.

The elements arranged on the conceptual structure map are language, image (still image or moving image), sound, environmental information (temperature and humidity, light and dark, color, contact, smell, taste, etc.), control program and the like. In particular, by arranging some control program, it is possible to control the behavior (movement, utterance, expression generation, etc.) of the robot that moves and acts autonomously.

If the self-learning function is prepared as a control program, not only the input from the outside (human side) but also the
The robot can autonomously learn.

With such a function, a robot having this conceptual structure map as a brain autonomously approaches an object of interest, avoids danger, etc., and can perform more efficient actions by a learning function. become. Even for a new event that has not been learned, it is possible to autonomously think and judge and perform some action by the association / creation function described above.

On the conceptual structure map, not only the knowledge information as described above but also emotion (kansei / preference) information can be arranged. Emotions have a role in motivating the maintenance and behavior of living individuals and species. Therefore, the robot having the conceptual structure map can be motivated based on its own interests and interests and can act autonomously without program control from the outside (human side).

Further, by arranging the self-replication function program on the conceptual structure map, it is possible to copy or back up the knowledge information and emotion information obtained by the user, and to damage the self-function by any accident or failure. Even if received, the function can be repaired autonomously.

The conceptual structure map held by the robot is
For example, it is also possible to copy information of a specific person or animal. With this function, the robot can be used as a tool for predicting and simulating the behavior of a specific person or animal.

In any of the functions and processes described above, the concept structure map is displayed in an audio-visual manner so that any concept can be obtained at present and the relationship between the concepts can be visualized and confirmed.

Therefore, a robot having the above-described functions can autonomously think, feel, memorize, process, and reflect on behavior, and can function more humanly than a conventional robot.

A more specific application example when used as the brain of the robot described above will be described below. (1) Pet Robot In recent years, robots (pet robots) as pet objects at home have been receiving attention. In each case, the purpose is to generate an emotional response such as a gesture that makes a person happy when stroked, or an angry gesture when struck.
However, a conventional pet robot causes a specific pattern of behavior according to a situation based on a database stored in advance.

By using the conceptual structure map formed according to the present invention, it is possible to cope with an unexpected situation, and to make a subtle change and expression of emotions, thereby realizing a robot with higher pettyness.

FIG. 16 shows an apparatus for generating a conceptual structure map stored in the pet robot. Note that, in FIG. 16, portions corresponding to those in FIG. The physical stimulus output means 12 gives the behavior of the robot as a physical stimulus. The behavior of this robot is
It may be provided by an image or by using the robot itself. From the autonomous input means 14, a user's linguistic response such as a call instruction for the behavior of the robot is input and input to the correction coefficient determination means 20.

The automatic measuring means 16 measures the user's gestures, facial expressions, how to make contact with the robot, etc. with respect to the behavior of the robot using sensors, and uses the measured data as the user's non-verbal response as a conceptual map generating means. Enter 18. The concept map generation means 18 generates an initial concept structure map representing the relationship between the behavior of the robot and the non-verbal response of the user. As a result, a conceptual structure map in which the user's sensibility and demand for the behavior of the robot are structured is generated.

The concept map automatic correction means 22 corrects the initial concept structure map with the correction coefficient determined by the correction coefficient determination means, and outputs the result from the result output means 24. As shown in FIG. 17, by storing the corrected conceptual structure map in the pet robot 70 and using it for generating the behavior of the robot, it is possible to cause the robot to perform a behavior suited to the user's sensitivity and demand. .

The robot 70 is provided with a sensory device 72 such as a tactile sense and an auditory sense, and an effect device for generating a behavior. The user uses the behavior of the robot as a physical stimulus, and inputs a linguistic response to the physical stimulus via the sensory device 72, thereby constantly correcting the correction coefficient for the behavior of the robot to generate an optimal conceptual structure map. Can be. (2) Nursing robots In nursing elderly and disabled people, not only physical support but also mental support is important. As in animal therapy, a technique for achieving mental stability by contacting small animals is known. The pet robot described above can be effectively used as a care robot in such care situations.

Also, especially in the case of elderly people, there is a problem of loss of heart due to separation from spouse. Since the concept structure map of the present invention can be used to configure a part of the concept structure of a spouse with the concept structure map, the robot storing the concept structure map in the brain is important as a talking partner or a consultant. It can be a mental support, so to speak, a mental welfare device.

The concept structure map of the nursing care robot can be created by the concept structure map generation device shown in FIG. (3) Friend robots (counseling robots) In recent years, crimes and problem behaviors due to mood changes such as "killing" and "mukatsuku" of young people have been attracting attention. One of the reasons for these moods is that they cannot fully objectively grasp their feelings and ways of thinking, and their ego identity is unclear. Since the conceptual structure map generated by the present invention can be visualized and audible, the user can objectively grasp his / her own way of thinking and feeling by looking at his own conceptual structure map.

[0099] Such objectification is usually performed on a daily basis in conversations with friends and the like. In the conversation,
They understand each other's way of thinking, make opinions and make proposals, and make objectives with friends. Therefore,
A friend robot (counseling robot) that has a brain that stores a conceptual structure map has the function to make the concept objective,
You can work like a friend or counselor.

FIG. 18 shows an apparatus for generating a conceptual structure map stored in the counseling robot. Note that FIG.
In the following description, parts corresponding to those in FIG. The physical stimulus output unit 12 gives the user an inquiry from the robot as a physical stimulus. From the autonomous input unit 14, a user's linguistic response such as a subject and a keyword is input and input to the concept map generation unit 18.

The automatic measuring means 16 measures the tone, speed, intonation, facial expression, complexion, heart rate, eyes, perspiration, etc. of the user's utterance in response to the robot's inquiry with a sensor, and measures the measurement data of the user in a non-verbal manner. The response is input to the correction coefficient determination means 18. Concept map generating means 18
Then, an initial conceptual structure map representing the relationship between the robot's question and the user's linguistic response is generated. As a result, a conceptual structure map is created in which the user's sensibility and request for the robot's inquiry are structured.

The concept map automatic correction means 22 corrects the initial concept structure map with the correction coefficient determined by the correction coefficient determination means 20, and outputs the result from the result output means 24. FIG.
As shown in FIG. 7, by storing the corrected conceptual structure map in the friend robot and using it for generating a question for the robot, it is possible to cause the robot to make a question suited to the user's sensitivity and demands. it can. (4) Robots for elucidating psychological functions Human psychological functions have not yet been completely elucidated. In particular,
Emotions, sensibilities, and preferences are the most elusive areas. The conceptual structure map obtained by the present embodiment visualizes and displays intangible elements such as concepts (languages, images, sounds, etc.) in a multidimensional space. This conceptual structure map can be prepared in advance, but can also be generated by automatic learning. Therefore, elements that cannot be prepared or considered by the researcher in advance can be obtained autonomously.
That is, it can be implemented as a psychological function simulation robot for elucidating a human psychological function or a psychological prediction robot.

Next, a fifth embodiment in which the conceptual structure map generated by the present invention is used as a conceptual structure map of an autonomous mobile unit will be described.

Since the conceptual structure map of the present invention can be generated with a low input cost for the user, it is incorporated into a device requiring human operation by utilizing this point, and the human operation is reduced. Suitable to do.

When the conceptual structure map according to the present embodiment is applied to an autonomous mobile body such as a car requiring human operation, environmental information, control information, and the like are determined based on actions performed by a user (driver). And a driver's preference relationship with respect to these pieces of information, and a conceptual structure map is generated. The conceptual structure map is used so as to produce a result in which later driving control, warning presentation, and the like are adapted to the driver's preference. Of course, the system automatically compensates for security. That is, the system can be prevented from running less secure models, no matter how dangerous driving is preferred.

Referring to FIG. 19, a description will be given of a conceptual structure map generating apparatus according to the present embodiment used in a situation-adaptive vehicle interior information providing system which is an automobile control apparatus. A physical stimulus generation unit (not shown) of the conceptual structure map generation device collects data on road conditions such as environmental information and information obtained from ITS. The physical stimulus output means 12 applies a physical stimulus to the driver using not only a display system such as a display but also all changes perceivable by the driver such as a reaction force of a steering wheel and a depressed weight of a brake. Is possible.

With respect to the information that can be perceived by the driver, the psychological state at the time when a certain sensation occurs, that is, the autonomous response and the non-autonomous response are detected by the autonomous input means 14 and the automatic measuring means 16, respectively. Immediately after the control portion of the system has performed some control, it is preferable to make corrections and control to minimize the trouble of the driver as much as possible.

In the autonomous input means 14, linguistic information is input by measuring an autonomous response, which is a conscious action of the driver, by a microphone, a switch, a button, or the like.
For example, a linguistic user's instruction such as "quickly holding" or "softer" is input in the sense of operation, and is input to the correction coefficient determining means 20 to correct the conceptual structure map for control. As a result, the control method can be adapted for each individual by reducing the play of the steering wheel. The adapted resulting control method is adjusted after judging safety based on subsequent user actions.

The automatic measuring means 16 captures, as data, patterns of gestures and actions recorded by a camera installed in the vehicle, blinks, eyes, face directions, arm directions, and the like. In addition, the time from when the environment changes until the operation is triggered, and physiological information (heart rate and palm blood flow) from various sensors attached to the steering wheel
And other data are also collected. In the case of carelessness ahead, a stronger safety control system is activated.

In this system, the physical stimulus output means 10
From the user, a driving situation (environmental situation, vehicle situation, etc.) obtained by data collection is given as a physical stimulus.

The automatic measuring means 16 includes operating characteristics (operating amounts of a steering wheel, an accelerator, a brake, a blinker, a navigation, etc., their operating timings, reaction speeds thereof, etc.), their physical and mental states (gaze, blink, heart rate, sweating). ), That is, the non-verbal response of the driver (the current driving characteristic of the driver) is detected and input to the concept map generating means 18. The concept map generating means 18 represents the relationship between the driving situation and the driver's response based on the driver's non-verbal response (the current driving characteristic of the driver) measured by the automatic measuring means 16, and displays the current driving characteristic. Generate a conceptual structure map that structures.

The autonomous input means 14 is a driver's linguistic response to a question or the like for detecting a target which is a physical stimulus from the physical stimulus output means 10 (driver's target).
Is input to the correction coefficient determining means 20. The driver's linguistic response may be a request for movement (such as being in a hurry, traveling on a wide road, etc.), a request for a vehicle (eg, increasing the distance between vehicles, or paying attention to the rear view, etc.) , Right leg has muscle pain etc.).

The automatic correction of the conceptual structure map by the automatic conceptual map correcting means 22 is performed in accordance with the correction coefficient determined from the driver's linguistic response input from the autonomous input means 14, whereby the conceptual structure map is converted. The target driving characteristics are corrected, and the distance between the elements is adjusted.
When the fatigue data is accumulated, such as during long-time operation, the measured value of the automatic measuring means 16 is input to the correction coefficient determining means 20, and the conceptual structure map is obtained based on the automatic measuring means 16 data. Is corrected. In other words, the conceptual structure map is adjusted based on the temporal integration value of both.

As shown in FIG. 20, the generated conceptual structure map is applied to optimization of information provision in the vehicle compartment.
It is possible to exchange at the will of the driver. Of course, it is possible to share and experience the operational feeling of friends while giving top priority to safety. In addition, when there is no adjusted personality data, such as when operating a car for the first time, a standard mode for general drivers is prepared as a conceptual structure map and estimated from data such as age, gender, driving history etc. A driving operation feeling may be provided.

Next, a more specific application example of the system using the conceptual structure map of the present embodiment for the autonomous moving body described above will be described below. (1) Vehicle control system tailored to the driver's personality The vehicle control program is created with the primary purpose of avoiding danger. In a normal driving situation, the driver is required to be familiar with the characteristics of the vehicle and perform an operation suitable for the vehicle. By storing the driving habits, preferences, and personalities of the driver as data in the present system, it becomes possible to control the vehicle suitable for the driver. Specifically, when changing lanes, the interval and timing at which the driver can change lanes with a margin can be predicted, and the predicted contents can be reported by changing the volume of the direction indicator and the like. It is also possible to adjust the inter-vehicle distance to suit the driver so that the driver can relax and operate. The preference of the steering wheel operation amount and the actual tire direction change amount (understeer or oversteer preference) can be adjusted for each individual. (2) Optimal Navigation System By using the conceptual structure map described above, it is possible to provide a comfortable route setting method for navigation. For example, it is also possible to provide the optimal driving course for the driver to the destination by searching for a route while keeping the scenery and speed feeling that the driver likes in the conceptual structure map is there. Further, even after the destination is set, it is possible to select whether to pass through a seaside road or a mountain road (forest road) according to the driver's preference. In addition, learn the driver's favorite scenery and driver's fatigue pattern to encourage breaks at appropriate places,
You can select your favorite tourist spot information and give advice. Of course, nothing can be presented to the driver who is bothered by the advice from the car. (3) Tension relief system In order to drive safely for the driver, it is preferable that the driver can drive in a relaxed state. For this reason, it is necessary to promptly introduce the user to a relaxed state when he / she feels irritated due to traffic congestion or reckless driving of another vehicle. The relaxation method selects one or more of the most optimal devices from among air conditioning, music, games, steering wheel operation, etc. according to the characteristics of the individual, and relaxes the driver. (4) Problem driving elucidation system Everyone is driving carefully at the time of a beginner driver. However, as the user gains experience and gets used to it, the probability of driving (careless driving) that can cause dangerous driving increases even if a serious accident does not occur. By using this system, by calculating the "safety degree" from environmental information (information on other things, etc.) and the driver's own condition, it is possible to inform the driver of dangerous driving according to the driver's wishes and to inform them of problems. Is possible. (5) In-vehicle environment operation support system As various devices are installed in a vehicle, attention may be lost from driving during operation of the devices. By inferring "the in-vehicle equipment that the driver wants" with this system, air conditioning adjustment, lighting of lights, etc. are performed automatically, so that the driver can concentrate on driving and reduce accidents. Examples of applications in other fields (1) Communication support system By seeing each other's visualized conceptual structure map for each individual, the issues and points of the other party can be seen at a glance, helping communication and preventing passing or misunderstanding. .

Since the conceptual structure map has a smaller amount of information than image information and audio information, the conceptual structure map information can be easily exchanged via a communication line. For example, in the case of presentations or presentations, if the claimed points are formed in advance as a conceptual structure map, important parts will be automatically enlarged, the volume of the speaker will be increased, and the inflection will be clear and the loop points will be clear. Can be presented.

Also, an interior advice system, a travel plan creation system (providing a plan for a desired journey), a system for making various design proposals such as clothes design and automobile design, and a health advice system (hair advice / beauty advice (desired atmosphere) Advice), sports gym advice, meal advice system (as a consultation for favorite ingredients and dishes), relaxation system (as the introduction of "healing" such as color therapy and aromatherapy), gift decision system (lover, family, etc.) (As a decision tool when a gift is given to a friend or important person), etc. (2) Market forecasting system For exchange and stock transactions, various statistical and mathematical information,
It depends on emotional information as well as political information.
For example, "that company is dangerous" or "this stock is interesting"
Transactions may be made for emotional reasons such as.

By expanding the conceptual structure map from personal use to collective use, for example, a conceptual structure map in which emotional information in a certain stock market is mapped can be generated. By embedding control programs such as buy and sell orders at the same time, it can be used as a tool for simulating the market movement.

In addition, it can be used as a selling tool for retail stores to improve the accuracy by cooperation with POS data such as convenience stores. (3) Preference database There is an attempt (media on demand) to provide favorite content in real time through cable TV or the Internet. The concept structure map can be used to search for content according to the user's preference.

In addition, as a compatibility diagnosis of a marriage introduction office, etc.,
This can be used as a team formation support for a project or the like (simulation of the optimal team configuration). (4) Comfort provision system By embedding the conceptual structure map in the information home appliance, for example, when "I want to eat curry today", the information network compatible refrigerator lists the materials necessary for the user's favorite curry. Can be up.

In addition, when performing driving support and route search in car navigation, it is possible to propose seas, mountains, and the like according to personal preferences.

Also, it can be used as a method for optimizing an information portal site such as the Internet.

[0123]

As described above, according to the present invention, a conceptual structure representing the relationship between a physical stimulus to a human and the first response of the human to the physical stimulus is changed as the physical stimulus is given to the human. Since the correction is made based on the second response generated, it is possible to obtain the effect that the input cost of the user himself can be reduced.

Further, if the concept structure is corrected by a non-verbal response which is an objective response, there is an effect that the expression accuracy can be improved and a concept structure map can be generated.

[Brief description of the drawings]

FIG. 1 is a block diagram of a first embodiment of the present invention.

FIG. 2 is a diagram showing a conceptual structure map showing a relationship between verbal response information and physical stimulus information according to the first embodiment.

FIG. 3 is a diagram illustrating a conceptual structure map corrected using non-verbal response information according to the first embodiment;

FIG. 4 is a block diagram of a second embodiment of the present invention.

FIG. 5 is a block diagram showing a conceptual structure map generation device of the emotion-coupled interior design image output system.

FIG. 6 is a flowchart illustrating a process of generating a conceptual structure map of the emotion-coupled interior design image output system.

FIG. 7 is a plan view showing a screen for inputting an impression word or the like for an evaluation image of the emotion-coupled interior design image output system.

FIG. 8 is a plan view showing a screen for generating an initial conceptual structure map of the emotion-coupled interior design image output system.

FIG. 9 is a plan view illustrating a process of correcting an initial conceptual structure map of the emotion-coupled interior design image output system with a language index, a behavior index, and a physiological index.

FIG. 10 is a diagram showing an example of a physiological index score of a subject.

FIG. 11 is a diagram illustrating an example of a physiological index score of a subject.

FIG. 12 is a diagram illustrating an example of a total score of a physiological index of a subject.

FIG. 13 is a diagram illustrating an example of a measurement result of a preference determination time.

FIG. 14 is a diagram showing an example of an individual concept structure map and a general concept structure map of the emotion-coupled interior design image output system.

FIG. 15 is a block diagram of an emotion-coupled interior design image output system.

FIG. 16 is a block diagram showing a conceptual structure map generation device for generating a conceptual structure map used for a pet robot or a care robot.

FIG. 17 is a block diagram of a robot such as a pet robot or a care robot provided with a conceptual structure map.

FIG. 18 is a block diagram showing a conceptual structure map generation device for generating a conceptual structure map used by a friend (counseling) robot.

FIG. 19 is a block diagram showing a conceptual structure map generation device for generating a conceptual structure map of the situation-adaptive vehicle interior information providing system.

FIG. 20 is a block diagram of a situation-adaptive vehicle interior information providing system.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Physical stimulus generation means 12 Physical stimulus output means 14 Autonomous input means 16 Automatic measurement means 18 Concept map generation means 20 Correction coefficient determination means 22 Concept map automatic correction means 24 Result output means

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Hiroyuki Furnishi 41-41, Chuchu-Yokomichi, Nagakute-cho, Aichi-gun, Aichi Prefecture Inside Toyota Central Research Laboratory Co., Ltd. 41, Yokomichi, Toyota Central Research Institute, Inc. (72) Inventor, Tetsuro Kurahashi 41, Chukuji Yokomichi, Nagakute-cho, Aichi-gun, Aichi Prefecture, Japan 1 Toyoda Central Research Institute, Inc. No. 41, No. 41, Chuchu, Nagakute-machi, Aichi-gun F-term in Toyota Central R & D Laboratories Co., Ltd. 5B075 ND34 NR02 5E501 AA02 AC23 AC34 BA03 BA15 CA08 CB02 CB05 CB09 CB14 CB15 EA21 FA14 FA32 FA44 FA46

Claims (13)

[Claims] 1. A conceptual structure representing a relationship between a physical stimulus to a human and a first response of the human to the physical stimulus is corrected based on a second response generated as the physical stimulus is applied to the human. And obtaining a conceptual structure adapted to the emotional structure of the individual. 2. The conceptual structure according to claim 1, wherein said first response is one of an autonomous response and a non-autonomous response, and said second response is the other of an autonomous response and a non-autonomous response. Method. 3. The method according to claim 2, wherein the non-autonomous response is at least one of a speech measurement value and a physiological measurement value. 4. The conceptual structure of claim 1, wherein said first response is one of a verbal response and a non-verbal response, and said second response is the other of a verbal and non-verbal response. Method. 5. The method according to claim 4, wherein the non-verbal response is at least one of a speech measurement and a physiological measurement. 6. The conceptual structure according to claim 1, further comprising: measuring environmental information around the human as the physical stimulus is applied to the human; and correcting based on the environmental information. Method. 7. A conceptual structure representing a relation between a physical stimulus to a human and a first response of the human to the physical stimulus is corrected based on a second response generated as the physical stimulus is applied to the human. And a concept structuring apparatus characterized by obtaining a concept structure adapted to an individual's emotional structure. 8. The conceptual structure according to claim 7, wherein said first response is one of an autonomous response and a non-autonomous response, and said second response is the other of an autonomous response and a non-autonomous response. Device. 9. The structuring apparatus according to claim 8, wherein the non-autonomous response is at least one of a speech measurement value and a physiological measurement value. 10. The conceptual structure according to claim 7, wherein said first response is one of a verbal response and a non-verbal response, and said second response is the other of a verbal and non-verbal response. Device. 11. The conceptual structuring apparatus according to claim 10, wherein said non-verbal response is at least one of a speech measurement value and a physiological measurement value. 12. The conceptual structure according to claim 7, further comprising: measuring environmental information around the human as the physical stimulus is applied to the human; and adding and correcting the environmental information. Device. 13. The concept structuring method according to any one of claims 1 to 5, or a concept structure obtained by the concept structuring apparatus according to any one of claims 7 to 11. apparatus.